The goal of this project is to maintain and improve the PKB software suite. PKB is an object-oriented data base containing structural data from the Protein Data Bank. It provides functions for retrieval of substructure data based on conformational and/or sequence motifs. PKB also provides functions for molecular-graphic and statistical analysis of protein substructures. Several new functions have been added to PKB. One set extracts information on non-standard covalent structure, such as bonds to prosthetic groups or metal ions, or amino acid residues which are cross-linked via such bonds. These functions are useful in querys concerning protein-prosthetic group interactions, and in controlling for the presence of such interactions in other querys. Another set of new functions applies crystallographic symmetry to generate lists of crystal packing contacts and corresponding symmetry operators. These functions are useful for examining the crystalline environments of different proteins, and inquerys pertaining to crystal packing interactions. Other new functions extract and compare protein sequence data, as distinct from crystallographic coordinates, and identify protein "core" substructures. These functions are useful in retrievals where structural attributes must be linked with entries insequence data banks. Many PKB functions were also modified to accommodate syntax changes in current versions of the S language, in which much of PKB is coded. PKB source has been made available to the community via electronic mail. The importance of this project is in providing software tools needed to create "views" of the Protein Data Bank that address specific biological questions. PKB is central to the work of several researchers addressing questions instructure analysis and prediction, at NIH and elsewhere. PKB also provides a means to create a new data base designed specifically for homology modeling.